12 years ago, William Kehler entered the alcohol industry with three brands he created, bootstrapped, and scaled. The last one, Bake Sale, which he describes as “Fireball-meets-crumble cookie in a shot,” was the home run.

Getting there wasn’t cheap. Kehler spent years working within alcohol’s three-tier distribution system, which he says is expensive for brands to navigate when they’re starting out and don’t have financial backing. Margins are thin, marketing spend is high, and the standard playbook of running promos in bars and liquor stores to build volume tends to put brands deep in the red before they ever climb out. Kehler says most brands end up on one of two paths: they lose money until they hit critical mass and get acquired, or they build a small tasting room and distillery operation, sell bottles at a decent margin, and struggle to scale beyond that.

The E-Commerce Pivot That Saved Bake Sale

The kickstarter for Bake Sale that changed its trajectory and made the brand take off was when alcohol e-commerce and D2C became legal.

Federal law sets nationwide baseline rules for alcohol products and shipments, while states decide whether direct-to-consumer shipping is allowed and impose their own licensing, volume, and delivery conditions. The 21st Amendment gives states significant authority over alcohol within their borders, but that authority is not absolute because federal law can still override conflicting state rules in some circumstances.

The patchwork is especially strict for spirits: many states still do not allow DTC spirits shipping at all, while others allow it only with licenses, volume caps, or other limits.

Basically, if a state allows a shipment, you still have to comply with federal requirements; if a state prohibits it, you generally cannot ship there.

Kehler seized the opportunity and quickly figured out how the model worked and where the levers were. He concentrated all of Bake Sale’s supply and funding into direct-to-consumer sales rather than splitting resources across channels. It worked. The brand scaled into six figures in sales, driven almost entirely by online orders of the dessert-flavored shot.

The switch to D2C sales catapulted growth ahead of operations so fast that they couldn’t keep up with demand because they were still packing every order by hand.

Building an Influencer Army

Kehler also credits part of Bake Sale’s D2C success to an in-house, AI-driven influencer program. Rather than going through influencer marketing agencies, where creators already know the value of their audience and price accordingly, Kehler’s team built a system to find and onboard micro-influencers who had never monetized their content before.

In exchange, those creators got product instead of a paycheck. Over the course of one year, the program recruited 3,000 influencers to post on Bake Sale’s behalf, generating 55 million impressions.

The model only works for a certain kind of product, Kehler says. “If you have a very generic product, this doesn’t work. You’re not going to get people that excited to try your product if it’s undifferentiated.” A novel product will attract attention, whereas a commodity item like plain vodka won’t.

Before AI-generated content became viable, Kehler’s team used the influencer footage itself as raw material for paid ads, testing large volumes of that content against the algorithm to see what performed. He says that approach is still one of the only ways to compete now that so many brands are running high volumes of content as ads. Without enough content in rotation, a brand can’t keep up.

Training AI for Ad Development

His team has developed a static ad engine that generates AI content and tests it at scale. He says while AI-generated imagery is getting more realistic, what’s usually missing is the emotional direction, the sense that an image captures who a customer wants to be or already believes they are. Kehler’s system is built to reverse-engineer a brand’s visual identity from its existing website and social content, then generate new images that match that identity rather than a generic AI look.

He points to a current client, a matcha energy drink brand, as an example. The brand’s positioning combines wellness and energy, categories that usually pull in opposite directions visually. Wellness content tends to read as calm, but an energy drink ad can’t look sleepy. Kehler’s team builds a model of how each brand speaks and looks, then generates content based on those parameters.

Kehler adds that his services and customized AI tools help brands sell more online, whether that means building a website, optimizing sales pipelines, or running the influencer and ad systems his team has already built.

Kehler pushes back on the idea that automating these tasks pushes people out of a job. He hopes that tools like these free people up to spend their time on the parts of work they’re actually good at and enjoy, sales calls and closing deals rather than the repetitive grind around them. The mindless, repetitive tasks, he says, are exactly the ones worth automating.

The Workaround Alcohol Brands Already Use

It’s easy to see how hemp beverages could be regulated under alcohol’s framework when it comes to D2C sales since they both already abide by strict regulations and maneuver a patchwork of state-by-state rules.

If that happens, Kehler points to a structure already in use in alcohol as a likely model: brands partner with a network of licensed retailers spread across multiple states, and each retailer fulfills orders locally rather than one warehouse shipping nationwide. A customer buying from a brand’s website appears to be ordering directly from that brand, but the order actually routes to whichever partner retailer is licensed to deliver in that customer’s state. States that don’t allow the shipment, Mississippi among them, get blocked automatically at checkout, with no friction on the customer’s end.

The challenge is that a brand needs retail partners in a particular state to cover its customer base, and retailers have to physically stock the product rather than order on demand. That structure tends to favor companies that already have big distribution networks. Mailing directly, by contrast, lets a small brand reach customers in multiple states without a retail footprint at all, which is how hemp brands have been able to reach customers who couldn’t get the product any other way.

Clearing Inventory Before the Clock Runs Out

Regardless of what happens with regulation at the end of the year, hemp brands are sitting on inventory that has an expiration date, and Kehler says that inventory needs to move soon. Retail expansion is slow even in normal conditions, and right now stores that already carry a product are cautious about adding anything new until there’s more regulatory clarity. D2C, by contrast, can be scaled up quickly without knocking on new doors. Kehler’s team is focused on helping brands move volume through that channel before time runs out, whether they’ve never touched e-commerce. He says his team can help a brand sell through a meaningful amount of inventory before that window closes.

Kehler also points to the data advantage of running an active D2C channel. A brand selling through retail alone often can’t tell whether 20 weekly sales in a store came from one bulk buyer or 20 different customers, information that’s effectively invisible without a direct sales channel. D2C generates the numbers investors actually want: cost to acquire a customer, the share of first-time buyers who come back, and what a dollar of ad spend reliably converts into. Kehler says many investors coming into hemp aren’t from the industry themselves. They’re numbers people, and a brand that can show clean acquisition and conversion data has an easier time making its case.